An experimental study into the default reading of constraint diagrams

Constraint diagrams (Kent, 1997) are a complex diagrammatic notation designed to express logical statements especially for use in software specification and reasoning. Not surprisingly, since this is an expressive language, there are some difficulties in reading the semantics of a diagram unambiguously. Some extra annotations (in the form of a reading tree) disambiguate the diagrams. However, this extra requirement (of drawing a reading tree) places a burden on the user. An attempt to remove the need for such a reading tree (or perhaps to automatically generate a reading tree, which could be altered by a user if they wished to) has been given via an algorithm to generate a default reading from the diagram. This algorithm is based on a number of principles - most of which are properties of the diagram. We wish to know whether these principles are intuitive and whether the default reading reflects a good proportion of users' intuitions, and we have performed a user-based study to test this.